KR101941225B1 - System, method and program for analyzing the crying sound of baby - Google Patents

Abstract

The present invention relates to a method and program for analyzing a baby crying sound.
A method for analyzing a baby crying sound according to an exemplary embodiment of the present invention includes: receiving a crying data related to a crying sound of a baby (S100); Acquiring only a specific length of time from the crying data to generate divided section data (S300); The computer extracting a plurality of sampling data (S500); Extracting the vowel interval data from the vowel interval data and the vowel interval data based on the waveform change or the waveform repetition in the time domain, and extracting vowel interval data corresponding to the extracted vowel data (S700); Extracting feature information from the vowel interval data (S900); And an emotion information determination step (S1100) of comparing the feature information with reference data to determine emotion information corresponding to the crying data.

Description

METHOD AND PROGRAM FOR ANALYZING THE CRYING SOUND OF BABY BACKGROUND OF THE INVENTION [0001]

The present invention relates to a method for analyzing a baby crying sound, and more particularly, to a method for analyzing a baby crying to analyze a current psychological state of a baby.

Generally speaking, infants who do not speak communicate their opinions to others with facial expressions, behaviors, and sounds. Particularly, it has been found that a typical pattern of a baby exists according to the current state of the baby, such as hunger, gas chorus, drowsiness, inconvenience, and the like. In the past, It is now possible to scientifically analyze the current state of the infant by analyzing the cry of the child.

In this regard, a service and a platform for analyzing the psychological state of a baby using a baby cry are provided, for example, a baby crying analyzer is being distributed on the market.

Korean Patent Publication No. 2003-0077489 Korean Patent No. 10-0337235

In the conventional baby crying analyzing apparatus, an economical burden has arisen for a user to purchase a separate apparatus, and there has been an inconvenience in carrying the apparatus at all times. That is, the caregiver of the baby is very cumbersome and inconvenient because it is necessary to carry the device even when going out, and when the device is not carried, the baby's current condition can not be judged.

In addition, the conventional baby crying analyzing apparatus has a problem that it can not transmit a specific baby's psychological state by classifying the infant's state according to a crying sound into a limited limited emotional state and displaying it as a light emitting unit of the apparatus. It is inconvenient to obtain additional information through a manual.

In addition, existing infant crying analyzing apparatuses and programs have a long time to acquire the baby crying result and present the analysis result, which makes it difficult for parents to cope with baby crying rapidly.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to analyze a baby's crying sound using a PC, a smart phone, a tablet PC, and the like, A method for analyzing a baby cry that can be performed, and a program.

The present invention also provides a method and program for analyzing a baby crying sound, in which parents can quickly cope with baby crying by increasing the accuracy while reducing the procedure required for analyzing baby crying sounds I would like to.

It is another object of the present invention to provide a baby crying sound analysis method capable of accurately analyzing various current states of infants by analyzing a baby's crying sound.

The technical objects of the present invention are not limited to the above-mentioned technical problems, and other technical subjects not mentioned can be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, there is provided a method for analyzing a baby crying sound, comprising: receiving crying data for receiving crying data related to a crying sound of a baby; The divided interval data being data of each divided interval obtained by dividing the crying data by a specific time interval, wherein the divided interval data is obtained by acquiring only a specific length of time from the crying data; Wherein the computer extracts a plurality of sampling data, the sampling data including one or more samples, the sampling being obtained by sampling at a predetermined extraction time interval in a time domain; Extracting the vowel interval data corresponding to the vowel interval data from the vowel interval data and the vowel interval data based on the waveform change or the waveform repetition in the time domain; Extracting feature information from the vowel interval data, wherein the feature information includes a shape of a loudness variation pattern or a repetition interval; And an emotion information determining step of comparing the feature information with reference data to determine emotion information corresponding to the crying data.

Another embodiment is characterized in that the analysis is performed in the time domain without converting the crying data into the frequency domain.

According to another embodiment of the present invention, in the sampling data extracting step, sampling is performed at an extraction time interval in the process of receiving the crying data in real time, and the dividing interval data generating step includes: And to acquire cry data of a time interval corresponding to the divided section.

According to another embodiment, the feature information extracting step may extract the appearance frequency of the specific vowel interval data or the pattern type contained in the at least one vowel interval data, Extracting specific reference data having a pattern form included in vowel interval data in the crying data from the reference data and determining emotion information corresponding to the reference data as a state of the infant, And is stored in the server as sampling data having a specific pattern form.

According to another embodiment of the present invention, the crying data reception step includes the steps of: correcting the pitch of the crying data based on the pitch of the reference data; And removing noise data included in the crying data.

According to another aspect of the present invention, there is provided a method for generating vowel section data, comprising the steps of: generating a first feature vector expressing the feature information based on feature information extracted from the vowel interval data; And compares the vector with a second feature vector of each reference data to calculate emotion information corresponding to the crying data.

According to another embodiment of the present invention, there is provided a method of analyzing an emotion information, comprising: confirming an analysis result based on feedback on the emotion information; And constructing big data by matching the crying data with the emotion information.

According to another embodiment of the present invention, in the extracting of the feature information, a step of generating conversion data excluding samples below a certain reference value in the sampling data, And the emotion information of the baby is calculated based on the feature information of the waveform.

A baby crying sound analysis program according to another embodiment of the present invention is combined with a computer which is a hardware, executes the aforementioned method for analyzing a baby crying sound, and is stored in a medium.

According to the present invention as described above, a user, that is, a protector of an infant, can install an application program on a PC, a smart phone, a tablet PC, and the like, and analyze the crying sound of the infant by using the application program. .

Further, it is possible to divide the baby crying into a predetermined section, analyze the vowel section included in the crying, and determine the emotion information corresponding to the baby's crying based on the characteristic of the section, Various cries can be reliably analyzed.

In addition, the present invention can accurately discriminate a baby state from a method of analyzing baby crying in the frequency domain. Embodiments of the present invention provide the effect of distinguishing the causes of five baby cries in the same pronunciation.

In addition, since the present invention can determine the emotional state of a baby immediately after a certain pattern is confirmed in a signal continuously obtained in the time domain, it is possible to perform a conventional baby crying analysis method (in particular, And performing analysis through frequency domain transformations).

1 is a flowchart showing a schematic flow of a baby crying sound analysis method according to an embodiment of the present invention.
FIG. 2 is a diagram illustrating an example of determining an emotional state by analyzing a baby's crying sound connected between components in performing a crying sound analysis in a server according to an exemplary embodiment of the present invention.
3 is an exemplary diagram for generating divided interval data in crying data according to an embodiment of the present invention.
4 is an exemplary diagram for extracting vowel interval data from sampling data according to an embodiment of the present invention.
5 is a graph of crying data for a baby cry in a hungry state according to an embodiment of the present invention.
6 is a graph of crying data for an uncomfortable baby cry according to an embodiment of the present invention.
7 is a graph of crying data for a sleeping baby cry according to an embodiment of the present invention.
8 is a diagram illustrating an example of distinguishing emotion information according to a crying pattern using an analysis method according to an embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS The advantages and features of the present invention and the manner of achieving them will become apparent with reference to the embodiments described in detail below with reference to the accompanying drawings. The present invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art. To fully disclose the scope of the invention to those skilled in the art, and the invention is only defined by the scope of the claims. Like reference numerals refer to like elements throughout the specification.

Unless defined otherwise, all terms (including technical and scientific terms) used herein may be used in a sense commonly understood by one of ordinary skill in the art to which this invention belongs. Also, commonly used predefined terms are not ideally or excessively interpreted unless explicitly defined otherwise.

The terminology used herein is for the purpose of illustrating embodiments and is not intended to be limiting of the present invention. In the present specification, the singular form includes plural forms unless otherwise specified in the specification. The terms " comprises " and / or " comprising " used in the specification do not exclude the presence or addition of one or more other elements in addition to the stated element.

The term " computer " as used herein encompasses various devices capable of performing computational processing to provide results to a user. For example, the computer may be a smart phone, a tablet PC, a cellular phone, a personal communication service phone (PCS phone), a synchronous / asynchronous A mobile terminal of IMT-2000 (International Mobile Telecommunication-2000), a Palm Personal Computer (PC), a personal digital assistant (PDA), and the like. Also, the computer may correspond to a server that receives a request from a client and performs information processing.

In this specification, the term 'client' refers to a device that performs a series of processes of exchanging data with a server, or a device that performs data processing on its own based on acquired data.

In this specification, 'server' means a device connected to one or more clients through a network. The network may be a wired or wireless Internet, or may be a core network integrated with a wired public network, a wireless mobile communication network, or a portable Internet, or may be a core network including a TCP / IP protocol and various services located thereon, ), Telnet, File Transfer Protocol (FTP), Domain Name System (DNS), Simple Mail Transfer Protocol (SMTP), and the like.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to the embodiments of the present invention, examples of which are illustrated in the accompanying drawings.

A conventional baby crying analyzing apparatus or method performs voice analysis by converting crying data over a specific length into a frequency domain for analysis. Specifically, a formant corresponding to the feature information is extracted in the frequency domain. A formant is a band in which the spectral energy of a sound source is concentrated, which is specific to a certain tone, and refers to an acid of energy generated in a certain frequency region due to resonance when a vowel or a voiced sound close thereto is analyzed by frequency . In other words, the conventional apparatus for analyzing baby crying sound extracts vowels and voices characteristic of voiced sounds by analyzing based on spectrum in a frequency domain of a sound source.

On the other hand, embodiments of the present invention perform analysis in the time domain without converting the crying data into the frequency domain. Specifically, embodiments of the present invention acquire sampling data for performing analysis from crying data in a time domain, and extract characteristic information in a time domain from the obtained sampling data to analyze an emotional state of a baby (or an infant).

1 is a flowchart of a method for analyzing a baby crying sound according to an embodiment of the present invention.

Referring to FIG. 1, a method for analyzing a baby crying sound according to an exemplary embodiment of the present invention includes: receiving a crying data related to a crying sound of a baby (S100); Acquiring only a specific length of time from the crying data to generate divided section data (S300); The computer extracting a plurality of sampling data (S500); Extracting the vowel interval data from the vowel interval data and the vowel interval data based on the waveform change or the waveform repetition in the time domain, and extracting vowel interval data corresponding to the extracted vowel data (S700); Extracting feature information from the vowel interval data (S900); And an emotion information determination step (S1100) of comparing the feature information with reference data to determine emotion information corresponding to the crying data. Hereinafter, a detailed description of each step will be described.

The computer receives crying data related to the baby's crying (S100; crying data receiving step). In one embodiment, as shown in FIG. 2, when the computer is a server, the server receives crying data related to the baby's crying sound from the client of the user who wants to grasp the baby's emotional state based on the baby crying sound. In another embodiment, if the computer is a client (i.e., if the client device performs its own culling data acquisition and analysis), the client may obtain crying data for performing the analysis itself, via a microphone provided in the client Acquire baby cry through wired or wireless communication from a separate device that can obtain it.

When a client acquires a baby cry for analysis by a computer (i.e., a server or a client), the client can download and install a separate application or application in which the baby cry analyzing method is implemented. When an application or an application program receives an analysis request (for example, a baby crying sound recording button operation on the application user interface) from a user (for example, a parent user who intends to grasp a baby emotion) . Specifically, the user can generate crying data by collecting a baby's crying sound using a microphone module provided in a client, for example, a smart phone, while the application or application program installed in the client is running.

In addition, when the server performs the process of analyzing the baby crying sound, the application acquires (e.g., records) the baby crying sound and transmits it to the server and receives the analysis result (i.e., emotion state information based on the baby crying sound) It can play a role of displaying on the screen. In addition, in order to reduce the size of data transmitted to the server, the application may also generate sampling data for a specific segment to be analyzed, as described later. In addition, when the client performs the baby crying analysis process, the application or program includes all of the modules for the entire process to perform the baby crying analysis inside.

According to another embodiment of the present invention, the crying data receiving step (SlOO) includes correcting the pitch of the crying data based on the pitch of the reference data. The pitch correction process may be a process of correcting the pitch of the obtained crying data with the stored reference data, that is, the data to be compared with the crying data. The baby's emotional state is analyzed by using various patterns of the baby's crying sound, for example, sound volume, pitch, vibration, repetition interval, etc., so that it is possible to perform a preprocessing operation for correcting the pitch to be compared with the reference data. Specifically, the computer obtains the average pitch from the cry included in the cry data to be analyzed. When the reference data is set to a specific pitch, the computer compares the pitch of the data to be analyzed with the pitch of the reference data to correct the pitch of the data to be analyzed.

 According to another embodiment of the present invention, the crying data receiving step (S100) includes removing noise data included in the crying data. The accuracy and reliability of the analysis can be improved by eliminating noise that is not related to the crying based on the baby's crying. That is, even if the crying is the same according to the surrounding environment of the client and the client device, it can have different noise and sound patterns, and the accuracy of the analysis can be improved by correcting the same. For example, a computer may use a filter (e.g., a Band-Pass Filter) to remove noise data that falls outside the range of the cry.

As shown in FIG. 3, the computer acquires only a specific length of time from the crying data to generate divided section data (S300: divided section data generating step). The divided section data is data of each divided section in which the crying data is divided at a specific time interval. Since the analysis is directly performed in the time domain without being transformed into the frequency domain, a short time unit (for example, 0.025 second) can be set as a unit of the divided section data without acquiring a sufficient time length for the frequency domain transformation. Specifically, the time length of the divided section is equal to or longer than a time unit in which the emotion state can be analyzed through the waveform comparison as one or more waveform patterns of the crying data for each emotion state are included in the time domain. That is, in this embodiment, the divided section data is set at an interval of 0.025 seconds, but the present invention is not limited thereto, and the divided section intervals may be determined at various reference intervals according to the user's environment and setting.

The computer extracts a plurality of sampling data (S500: Sampling data extracting step). The sampling data includes at least one sample, and the sample is obtained by sampling at a predetermined extraction time interval in the time domain. For example, a computer can acquire sampling data at SampleRate 8000Hz, PCM signed 16bit.

In one embodiment, the sampling data extracting step (S500) may be performed before the divided section data generating step (S300) or after the divided section data generating step (S300). For example, when the crying data is acquired, the sampling data extraction step (S500) is immediately performed to generate the sampling data for sampling the entire range of crying data to be obtained, divide the sampling data into time units of a specific division period, And can be generated as divided section data. In addition, for example, the computer may execute the sampling data extracting step S500 only for a specific number of divided section data after generating the divided section data for the specific division section on the crying data.

In another embodiment, the sampling data extracting step (S500) may be performed simultaneously with the dividing section data generating step (S300). Specifically, the computer performs sampling immediately at the extraction time interval (S500; sampling data extraction step) in the process of acquiring or receiving the crying data in real time, and crying at a time interval corresponding to the division interval from the crying data acquisition requesting time And acquires one or more divided section data in a sampled form (S300: division section data generation step). When the computer acquires the crying data, the time required to generate the analysis target data for analyzing the baby crying sound is remarkably reduced by dividing the crying data at the same time in units of divisional sections and performing sampling (that is, performing the steps S100, S300, and S500 simultaneously) Can be reduced. Particularly, when a user terminal (for example, a smartphone in which a baby crying application is installed) analyzes baby crying without communication with a server, division data generation and sampling Data generation can be performed simultaneously.

The computer distinguishes vowel interval data from vowel interval data based on waveform changes or waveform repetition in the time domain, and extracts vowel interval data corresponding to the extracted vowel data (S700). Within the crying data where sampling is performed, a waveform change occurs between the beginning and end of a syllable. At the beginning of the syllable, consonants are included, but at the end of the syllable, the pronunciation of the consonants is excluded, leaving only vowel pronunciations. Therefore, a section including a consonant sound (hereinafter referred to as a consonant section) and a section including only a vowel sound (hereinafter referred to as a vowel section) have different waveform patterns. The consonant section appears first, the vowel section appears, and the section including the consonant pronunciation does not have a certain pattern, but the vowel section has the same pattern within the entire range. For example, as shown in FIG. 4, when the baby cries 'I', the section including the 'b' sound corresponding to the consonant has a non-uniform pattern, whereas only the vowel 'a' The included section shows a certain pattern, but shows a different pattern depending on the emotional state.

Thus, the computer distinguishes consonant intervals and vowel intervals (e.g., recognizes boundaries between consonant intervals and vowel intervals) based on waveform changes or waveform repetitions. Specifically, the computer recognizes the point at which the waveform changes to a section where the same waveform repeatedly appears (i.e., vowel section) in a section where the waveform does not show a certain pattern (i.e., consonant section) as a boundary between the consonant section and vowel section . Then, the computer extracts the vowel interval data corresponding to the vowel section after the boundary point between the consonant section and the vowel section. The vowel interval data is sampled data in a divided interval unit corresponding to vowel pronunciation.

In another embodiment, instead of extracting the vowel interval data after dividing the crying data at intervals of predetermined intervals, the computer may include a vowel section in a form that is easy to perform analysis after extracting vowel sections included in crying data A plurality of divided section data can be generated. That is, since the baby crying pattern occurs mostly in the vowel section, the computer may extract the vowel section included in the crying data first, and may generate the sampling data by sampling only the vowel section, instead of sampling the entire crying data.

The computer extracts feature information from the vowel section data (S900: feature information extraction step). The feature information includes a form of a loudness variation pattern or a repeated section. In one embodiment, the characteristic information extracting step S900 extracts the appearance frequency of the specific vowel interval data or the pattern type contained in the at least one vowel interval data. For example, as shown in FIG. 8, the computer extracts a specific pattern shape repeatedly appearing in the vowel interval data.

In addition, the baby may have multiple emotional states. In this case, one pattern form does not appear in one or more vowel section data but a plurality of waveform patterns can appear in the vowel section data, and the computer can extract a plurality of waveform patterns contained in one or more vowel section data.

Further, the computer can calculate the appearance frequency of each pattern in one or more vowel interval data. The computer can calculate the accurate emotional state of the baby by reflecting the occurrence frequency of each waveform pattern. Specifically, since the computer may have a plurality of vowel patterns within a specific time range (for example, the voices of the vowel pattern may be combined due to an increase in crying due to hunger) The number of times or frequency of appearance can be taken into consideration to grasp the situation of the baby.

5 to 7, the characteristic information extracting step S700 may further include generating transformed data of the vowel interval data around a specific reference value. Namely, in order to perform a quick and accurate analysis of the baby crying, the computer converts the sampling data (i.e., Figs. 5A, 6A and 7A) 5 (b), 6 (b), and 7 (b)). For example, a computer can process a sample larger than a certain reference value by 1 and process a sample smaller than a reference value to 0 to generate a digital signal.

Further, the computer can generate a plurality of level values by setting a plurality of reference values. For example, the computer may divide the entire range of sample values that may appear in the baby crying into a plurality of intervals, and determine a sample included in a specific interval as a level corresponding to the interval. Accordingly, the computer can generate the sampling data in a natural number pattern corresponding to the level of each section.

Accordingly, the computer can generate the digital data of the sampling data of each vowel section, and compare the digital data of the reference data with the digital data of the reference data to determine the baby's emotional state.

5 to 7, the computer processes the value itself or the sample in which the absolute value is smaller than the specific reference value as 0, and processes the value itself or the value whose absolute value is equal to or greater than the reference value as an absolute value The sampling data can be processed in such a manner as to be left behind. For example, since the frequencies and patterns that occur with a high degree of sound are the determining factors of the distinction of the baby's emotional state, the computer generates the conversion data by leaving only the samples over a certain reference value.

In another embodiment of the feature information extracting method, the computer may generate feature information of the vowel interval data in the form of a feature vector. That is, the feature information extracting step S900 includes generating a first feature vector expressing the feature information based on the feature information extracted from the vowel interval data. The sampled vowel interval data can have respective sounding widths, pitches, sound vibrations, repetition intervals, etc. according to sampling intervals, and can be extracted by feature vectorization. In other words, the feature vector may be a vector including one or more of the following characteristics of sounding, pitch, sound vibration, and repeat section of each sampling data.

The computer compares the feature information with reference data to determine emotion information corresponding to the crying data (S1100: emotion information determination step). The emotion information is information that embodies various current states of the baby, and may be information on hunger, inconvenience, sleepiness, comfort, gas charm, and the like. As shown in FIG. 10, the computer stores reference data for each emotion state of a baby or a baby, and compares the feature information of the reference data with the feature information extracted from the analysis target data to grasp the emotional state of the baby or the infant. The reference data is stored in the server as sampling data having a specific pattern form according to each emotion information.

For example, in the emotion information determination step S1100, specific reference data having a pattern form included in vowel interval data in crying data among at least one reference data stored in a computer is extracted, and emotion information corresponding to the reference data To determine the state of the infant.

Specifically, when the computer is a user terminal (for example, a smartphone or a tablet PC in which a baby crying analysis application is installed), the computer displays a collection interval sampled based on the reference data, And calculates emotion information corresponding to the data. For example, reference data (pattern data corresponding to each situation) for five situations (hunger, discomfort, sleepiness, trimming, gas true) are stored in a baby crying sound analysis application installed in a user terminal, It is determined which reference data is closest to the reference data and the baby's emotional state is determined.

When the feature information is extracted in the feature information extracting step (S900) to generate a feature vector, the emotion information determining step (S1100) compares the first feature vector with a second feature vector of each reference data, And calculates emotion information corresponding to the data.

When the feature information is extracted in the feature information extracting step S900 to generate digital data, the emotion information determining step S1100 may include a step of extracting a plurality of pieces of digital data (i.e., (I.e., first digital data) generated in the vowel interval data in the obtained crying data by comparing the first digital data with the second digital data, The baby's emotional state is calculated.

Referring to FIG. 8, an embodiment for distinguishing emotion information according to a crying pattern using the baby crying analysis method according to the present embodiment will be described in detail.

Specifically, FIG. 8A is an example of a vowel section pattern in a hungry state, FIG. 10B is an example of a vowel section pattern in a sleeping state, and FIG. 10C is an example of a vowel section pattern in an uncomfortable state. As shown in the figure, the vowel sections included in each cry were analyzed for vowel patterns included during 6.88 to 6.89 seconds of (a), 8.41 to 8.42 seconds of (b), and 1.83 to 1.84 seconds of (c) As a result, since it appears repeatedly in a relatively short interval of 0.01 second, it is possible to provide fast and reliable analysis results using short crying data compared with the existing baby crying analysis method.

In addition, since each of the emotion states has a clearly different pattern in the time domain, it is possible to guide the correct baby emotion state so that the parent performs appropriate coping. Specifically, the emotional state of the baby that the parent has to cope with can be classified into five types: drowsiness, discomfort, hunger, trimming, and gas charm. As the computer directs the five types of baby emotional states individually (i.e., the baby has only one emotional state) or complex (i.e., the baby is crying for several reasons), the user Sufficient response is possible.

The waveform pattern of each baby's emotional state will be described in detail with reference to Figs. 5 to 7. Fig. 5 (a), 6 (a) and 7 (a) are graphs of vowel interval data sampled for each emotion state, and FIGS. 5 (b), 6 (b) and 7 Is a graph of transformed data in which only the value of a certain reference value or more is left out of the samples of the vowel interval data, and the remaining samples are excluded (i.

When the baby's emotional state corresponds to "hunger", as shown in FIG. 5, the high and low notes are spaced narrow and show a regular pattern (eg, a 'a' sound pattern that sucks the milk). In addition, since 'gas true' and 'hunger' are similar baby states, they have a similar waveform pattern form in a plurality of crying data used for generating reference data.

When the baby's emotional state corresponds to 'discomfort', as shown in FIG. 6, the vowel sound 'a' appears relatively repeatedly and coarsely. In other words, because it is inconvenient, it gives a strong sound, and the high sound interval appears relatively thick, and the low sound interval is 2 ~ 4 times wider than the high sound interval. Baby cries collected in uncomfortable conditions are similar to hungry vowel patterns, but are more dense and irregularly formed between each pattern.

When the baby's emotional state corresponds to 'sleepiness', as shown in FIG. 7, the sound of the vowel 'A' is cut off like a bouncing sound, and the sound is stretched, and the sound is blocked and the sound is repeated. As the baby is drowsy, the neck is not able to make a sound because of its strength, so the interval of low notes is wide for a short interval and the interval of high notes is short.

Still another embodiment further includes generating reference data for baby crying analysis. In one embodiment of generating reference data, the computer may generate reference data by receiving characteristic information on a specific cry from the expert. For example, the server collects emotion state information about baby cries of various patterns from an expert, matches the emotion information corresponding to each crying pattern in advance, and manages the emotion information in a database.

Further, in another embodiment of generating reference data, when the computer is a user terminal (e.g., a terminal equipped with a baby crysis analysis application), the computer periodically or at certain times updates the reference data from the server, And stores the data. Since the user terminal receives only the reference data from the server and updates it, the user terminal does not need to communicate with the server at the time of analyzing the baby crying sound, so the data usage can be reduced and the time required for communication with the server can be reduced.

In another embodiment of generating reference data, feedback from the user about the emotion information determined based on the baby crying sound is received through the client, and the error of the emotion information provided as the analysis result is determined to supplement the reference data . That is, it is possible to check the analysis result based on the feedback on the emotion information. And constructing big data by matching the crying data with the emotion information.

The computer provides the analysis result (i.e., the calculated emotion information) to the user, and based on this, the result after responding to the baby cry can be inputted through the client. For example, the user may be asked whether the analysis result is effective, and if the positive feedback is received, it is determined that the analysis result is correct, and if the negative feedback is received, the analysis result may be determined to be in error. In addition, in the case of receiving negative feedback, the computer may be provided with a reward for the user and input data on the actual condition of the baby (i.e., the actual reason for the baby's crying or the actual crying of the baby).

This allows the computer to accurately match the baby cry pattern to the actual crying cause. Computers can improve the accuracy of analysis by generating accurate cry patterns for each emotion type as reference data through analysis of precisely matched cry patterns and causes of crying.

Further, in another embodiment, the computer grasps if it is not a pattern of baby crying (e.g., when the parent's voice is acquired together) and provides a baby crying request to remove the noise. That is, the computer can instruct the user that the baby crying has not been properly obtained, and request to reacquire the baby crying if the interval that matches the pattern included as the reference data is not confirmed.

As described above, the method for analyzing the baby crying sound according to an embodiment of the present invention may be implemented as a program (or an application) to be executed in combination with a computer as a hardware and stored in a medium.

The above-described program may be stored in a computer-readable medium such as C, C ++, JAVA, machine language, or the like that can be read by the processor (CPU) of the computer through the device interface of the computer, And may include a code encoded in a computer language of the computer. Such code may include a functional code related to a function or the like that defines necessary functions for executing the above methods, and includes a control code related to an execution procedure necessary for the processor of the computer to execute the functions in a predetermined procedure can do. Further, such code may further include memory reference related code as to whether the additional information or media needed to cause the processor of the computer to execute the functions should be referred to at any location (address) of the internal or external memory of the computer have. Also, when the processor of the computer needs to communicate with any other computer or server that is remote to execute the functions, the code may be communicated to any other computer or server remotely using the communication module of the computer A communication-related code for determining whether to communicate, what information or media should be transmitted or received during communication, and the like.

The medium to be stored is not a medium for storing data for a short time such as a register, a cache, a memory, etc., but means a medium that semi-permanently stores data and is capable of being read by a device. Specifically, examples of the medium to be stored include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage, and the like, but are not limited thereto. That is, the program may be stored in various recording media on various servers to which the computer can access, or on various recording media on the user's computer. In addition, the medium may be distributed to a network-connected computer system so that computer-readable codes may be stored in a distributed manner.

The method according to this embodiment may be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays , A controller, a microcontroller, a microprocessor, or the like.

While the present invention has been described in connection with what is presently considered to be practical exemplary embodiments, it is to be understood that the invention is not limited to the disclosed embodiments, but, on the contrary, You will understand. It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive.

10: Server
20: Client device

Claims (9)

A crying data receiving step in which the computer receives crying data related to a cry of a baby;
Wherein the divided interval data is obtained by obtaining only a specific length of time from the crying data, wherein the divided interval data is data of each divided interval in which the crying data is divided at specific time intervals, A division interval data generation step of generating a division interval data having at least one waveform pattern of data and analyzing an emotion state through waveform comparison;
Wherein the computer extracts a plurality of sampling data, the sampling data including one or more samples, the sampling being obtained by sampling at a predetermined extraction time interval in a time domain;
Extracting the vowel interval data corresponding to the vowel interval data from the vowel interval data and the vowel interval data based on the waveform change or the waveform repetition in the time domain, Extracting the vowel interval data as a vowel interval data when a waveform change occurs in one syllable in the crying data performed, the interval in which the same waveform appears repeatedly as a second half of the waveform change portion, is extracted as vowel interval data;
Extracting feature information from the vowel interval data, wherein the feature information includes a shape of a loudness variation pattern or a repetition interval; And
And an emotion information determining step of comparing the feature information with reference data to determine emotion information corresponding to the crying data,
The division section data generation step, the sampling data extraction step, and the vowel section data extraction step are performed in the time domain. The computer obtains crying data, divides the data into division sections and performs sampling, The analysis is performed in the time domain without converting the crying data into the frequency domain,
The feature information extracting step includes:
Extracting a frequency of occurrence of specific vowel interval data or a pattern type contained in the at least one vowel interval data,
When a plurality of waveform patterns appear in the vowel interval data in the characteristic information extracting step,
The feature information extracting step extracts the feature information by reflecting the occurrence frequency of each waveform pattern, and then determines the combined emotion state in the emotion information determining step.
Wherein each of the waveform patterns represents each individual emotional state constituting the combined emotional state,
In the emotion information determination step,
The reference data is stored for each emotion state of hunger, discomfort, sleepiness, trimming, and gas true according to the extracted feature information,
The feature information is compared with the reference data to determine the information as emotion information of hunger, inconvenience, sleepiness, trimming,
When the value of the sound height in one period is equal to or greater than the predetermined sound height, the interval is a high sound interval, and when the value of the sound height is less than the predetermined sound height,
Wherein the waveform pattern in which the interval between the high-frequency interval and the low-frequency interval is equal to or less than a predetermined interval in the one period corresponds to a waveform pattern of hunger or discomfort,
Wherein the waveform pattern when the length of the low-frequency section in the one period among the waveform patterns of the hunger or the inconvenience is not less than two times and not more than four times the length of the high-frequency section is determined as the feeling state of the inconvenience.
Baby crying analysis method. delete The method according to claim 1,
The sampling data extracting step may include:
In the process of receiving the crying data in real time, sampling is performed at an extraction time interval,
Wherein the division section data generation step comprises:
And obtaining crying data at a time interval corresponding to the divided section from the time of requesting the crying data acquisition. The method according to claim 1,
Wherein the emotion information determination step comprises:
Extracting specific reference data having a pattern form included in vowel interval data in the crying data from one or more reference data stored in the computer and determining emotion information corresponding to the reference data as a state of a child, And is stored in a server as sampling data having a specific pattern form according to each emotion information. The method according to claim 1,
The crying data receiving step includes:
Correcting the pitch of the crying data based on the pitch of the reference data; And
And removing noise data included in the crying data. The method according to claim 1,
And generating a first feature vector representing the feature information based on the feature information extracted from the vowel interval data,
Wherein the emotion information determination step comprises:
And comparing the first feature vector with a second feature vector of each reference data to calculate emotion information corresponding to the crying data. The method according to claim 1,
Confirming an analysis result based on feedback on the emotion information; And
And constructing big data by matching the crying data with the emotion information. The method according to claim 1,
The feature information extracting step includes:
And generating transformed data from the sampling data excluding samples below a certain reference value,
Wherein the emotion information determination step comprises:
And calculating the emotion information of the baby on the basis of the feature information of the waveform included in the conversion data. 8. A baby crying analysis program stored on a medium for performing the method of any one of claims 1 to 9 in combination with a computer which is hardware.

Images